The ideal way to to copy a git repository is using git itself. It provides features to efficiently track and download additions and updates, without requiring custom code to examine last modified times or forcing a redownload of all files. However, this normally means copying every file in the repository, including historical versions.
Fortunately, git provides a feature known as a sparse checkout. Combined with other git filtering features, this can be used to avoid a full download of the repository. You can configure it to only download and save specific directories or files to your computer. If you are only interested in a subset of the files provided here, this is the most efficient way to download them and keep them up to date.
Note
The instructions provided here will use the git command line. If you use a graphical tool to work with git repositories, you will need to determine whether and how it supports the sparse-checkout feature.
Start by cloning the repository using the following flags/options:
git clone --filter=blob:none --no-checkout --depth=1 https://github.com/Fernando2603/AzurLane.git AzurLaneENResources
Note
You are free to use a different directory name than AzurLaneENResources.
Each option has a specific purpose:
--filter=blob:none: Delays downloading the file's contents until it is actually checked out to disk.--no-checkout: Causes git to only create its special .git folder for managing the commit tree. This is essential to prevent git from needing to download all files.--depth=1(optional): Only download the most recent commit, omitting the history of changes. If you are interested in the history of how the asset files have changed over time, omit this option.
See git-clone for more details.
Its output will look similar to this:
Cloning into 'AzurLaneENResources'...
remote: Enumerating objects: 4047, done.
remote: Counting objects: 100% (4047/4047), done.
remote: Compressing objects: 100% (4047/4047), done.
remote: Total 4047 (delta 0), reused 2080 (delta 0), pack-reused 0 (from 0)
Receiving objects: 100% (4047/4047), 1.64 MiB | 3.82 MiB/s, done.
(The exact numbers in the output will likely be different.)
Next, configure the repository to use sparse checkout:
cd AzurLaneENResources
git sparse-checkout init
If you are using git bash from Git for Windows, you will see the branch name change from
(main) to (main|SPARSE) now.
Now we can check out the repository:
git checkout
Its output looks similar to this:
remote: Enumerating objects: 27, done.
remote: Counting objects: 100% (27/27), done.
remote: Compressing objects: 100% (22/22), done.
remote: Total 27 (delta 13), reused 9 (delta 5), pack-reused 0 (from 0)
Receiving objects: 100% (27/27), 3.66 MiB | 4.23 MiB/s, done.
Resolving deltas: 100% (13/13), done.
Updating files: 100% (31/31), done.
Your branch is up to date with 'origin/main'.
By default, a sparse checkout will only save files at the very top directory of the repository. In this repository, that includes only the top level JSON files and the README.
To actually download resource files, we need to add locations to the sparse checkout's list. The
add command is the main way to do so. For example, if you want to checkout the music files, you
can add the audio/bgm directory:
git sparse-checkout add audio/bgm
This will checkout only the files inside the audio/bgm directory. It may take a few minutes.
You can use the command again to add an additional directory:
git sparse-checkout add images/equipment
Note
You cannot specify a single file. You must use directories:
$ git sparse-checkout add images/equipment/1.png
fatal: 'images/equipment/1.png' is not a directory; to treat it as a directory anyway, rerun with --skip-checks
If you want to see the list of directories you've added, use the list command:
git sparse-checkout list
Output:
audio/bgm
images/equipment
If you would like images or audio from a specific ship or skin, consult ship_skin.json to locate
the ID number, and then add its folder to the sparse checkout. For example, Cassin's retrofit
skin has the ID 101039, so we can add it using:
git sparse-checkout add images/skin/101039 audio/voiceline/101039
(Notice that this will only check out the voice lines specific to that skin; the base voiceline are in the default skin's directory.)
To obtain updates from the repository, you should avoid git pull and git merge. These will not
respect your sparse checkout settings because they will attempt to perform a merge. Instead, use
fetch with --filter to obtain new commits:
git fetch --filter=blob:none
And then check out the newest files using git reset:
git reset --hard origin/main
Caution
Do not use reset if you have made any local changes to the repository's files or commits. It
will discard any uncommitted local file modifications and set the current branch to the most
recent remote commit. Resetting the branch's head can result in losing commits that have not
been pushed or included in another branch. (There are ways to recover such lost commits, but
it's better to avoid needing to in the first place.)
Use git merge instead if you have modified the repository. Even though it downloads more files,
you will still only be downloading recently added or changed files.
If you use set, it will completely replace the existing list of checkout directories:
git sparse-checkout set audio/bgm images/skin/101039 audio/voiceline/101039
Since this does not include images/equipment, it will remove that directory from the list
and delete the directory and files inside it from disk.
Invoking the list command will show that it is gone:
$ git sparse-checkout list
Output:
audio/bgm
audio/voiceline/101039
images/skin/101039
The following command replaces Cassin retrofit's skin and voiceline with her default skin and lines:
git sparse-checkout set audio/bgm images/skin/101030 audio/voiceline/101030
Important
Using set this way is current the only way to remove a directory from sparse-checkout.
If you have a long list of directories, it can be inconvenient to manually build up a complete set
command. To save effort, you can use the following bash command to remove that specific directory:
git sparse-checkout set $(git sparse-checkout list | grep -v -F 'images/equipment')
This command generates the current list using git sparse-checkout list, filters out the item
'images/equipment' (the -v means include items that do not match, and the -F disables
regular expressions), and then passes the entire list (combined into a single line) to the set
command.
You can use echo to verify the list before actually running the command:
echo $(git sparse-checkout list | grep -v -F 'images/equipment')
If you are familiar with bash and the tools provided alongside it, feel free to use that knowledge.
Last, calling set with no parameters will clear the entire list:
git sparse-checkout set
This will setting the repository back to the initial configuration, removing all files except the ones at the repository's top directory.
Important
Removing files from the sparse checkout list does not remove their data stored within the repository's hidden .git directory. They still take up disk space there. The only way to eliminate them from your disk entirely is to start with a fresh repository.
If you accidentally cause git to download a lot of files, want to remove a lot of locations, or
think that updates are making it large, feel free to just delete the repository directory
and create a new one. Git is entirely based on file manipulation and stores all its data inside
a hidden directory named .git. Since there is no centrally stored data for the repository,
it is completely safe to delete it the normal way for your system.